Using OSS models to save on inference costs without cutting quality

Braintrust now lets users call GLM-5.2 directly in playgrounds, prompts, and scorers with no configuration through July 31, letting teams experiment with open-source models to cut inference costs without sacrificing quality.